v0.6.0

December 2023

Highlights

• A deprecation policy is now in place for aeon and its subpackages.

• New classification algorithms are available in RED CoMETS and LITETime.

• A clustering subpackage for deep learning clustering has been added.

• A new experimental similarity search subpackage has been added, and the annotation module has been split into segmentation and anomaly detection.

• Minkowski distance has been added to distances.

• Lots of small improvements and bug fixes!

Anomaly Detection

Other

• [ENH] refactor annotation into segmentation and anomaly detection (#782) @TonyBagnall

Benchmarking

Bug Fixes

• [ENH,BUG] CD plot function has been simplified, fixing some errors. (#902) @dguijo

• [BUG, ENH] Fixes several issues and enhances plotting module (#950) @dguijo

Enhancements

• [ENH,BUG] CD plot function has been simplified, fixing some errors. (#902) @dguijo

• [ENH] remove two calls to convert_to (#947) @TonyBagnall

• [BUG, ENH] Fixes several issues and enhances plotting module (#950) @dguijo

Classification

Bug Fixes

• [BUG] Correct pipeline test fails (#855) @TonyBagnall

• [BUG] Callables for KNN (#967) @TonyBagnall

Documentation

• [DOC] Enhance distance function related docs (#969) @TonyBagnall

Enhancements

• [ENH] Adds RED CoMETS classifier (#779) @zy18811

• [ENH] Shapelet transform changes (#857) @TonyBagnall

• [ENH] change collection conversions from convert_to to convert_collection (#838) @TonyBagnall

• [ENH] Support unequal length for rdst (#891) @baraline

• [ENH] Remove case sensitivity from Rocket transform argument (#911) @TonyBagnall

• [ENH] Adding LITETime for Deep Learning classifiers (#923) @hadifawaz1999

• [ENH] DL Refactor (#937) @TonyBagnall

• [ENH] Swap transformer used in SummaryClassifier (#978) @MatthewMiddlehurst

Other

• [ENH] Adapt ShapeletTransform to work with unequal length (#820) @TonyBagnall

Clustering

Bug Fixes

• [ENH] Allow callables in TimeSeriesKMedoids (#925) @TonyBagnall

Documentation

• [DOC] Kmeans updated default for averaging method (#880) @chrisholder

• [DOC] Enhance distance function related docs (#969) @TonyBagnall

Enhancements

• [ENH] Deep Clustering module basis with an auto encoder FCN based example (#697) @hadifawaz1999

• [ENH] Allow callables in TimeSeriesKMedoids (#925) @TonyBagnall

• [ENH] Make clustering files private (#943) @TonyBagnall

Maintenance

• [MNT] Reduce test params for AEFCNClusterer (#868) @hadifawaz1999

• [MNT] Replacing BA by Mean in AEFCNClusterer test params (#879) @hadifawaz1999

Datasets

Bug Fixes

• [BUG] Ignore extra whitespace in loader (#884) @TonyBagnall

• [ENH] Add timestamped data loading for collections (#887) @TonyBagnall

• [BUG] Check if multiindex store unequal length series (#928) @TonyBagnall

• [BUG] test forecasting loader return meta data (#972) @TonyBagnall

Enhancements

• [ENH] Error message for incorrect unequal length multivariate input (#867) @TonyBagnall

• [ENH] Improve coverage for datasets module (#888) @TonyBagnall

• [ENH] Add timestamped data loading for collections (#887) @TonyBagnall

• [ENH] Add download_dataset function to _data_loaders.py (#906) @PaulRabich

• [ENH] Results loaders for published results (#890) @TonyBagnall

• [ENH] default loaders to not return meta data (#953) @TonyBagnall

• [ENH] Tidy up and restructure load to nested univariate (#987) @TonyBagnall

Distances

Bug Fixes

• [ENH] Allow callables in TimeSeriesKMedoids (#925) @TonyBagnall

• [BUG] Callables for KNN (#967) @TonyBagnall

Documentation

• [DOC] Enhance distance function related docs (#969) @TonyBagnall

Enhancements

• [ENH] Allow callables in TimeSeriesKMedoids (#925) @TonyBagnall

• [ENH] Minkowski distance (#904) @akshatvishu

Networks

Enhancements

• [ENH] Adding LITETime for Deep Learning classifiers (#923) @hadifawaz1999

• [ENH] DL Refactor (#937) @TonyBagnall

Maintenance

• [MNT] Default InceptionTime to use original model and not H version (#921) @hadifawaz1999

• [MNT] Update custom filters in H-InceptionTime and LITETime ot work with new tf versions (#949) @hadifawaz1999

Regression

Bug Fixes

• [BUG] Correct pipeline test fails (#855) @TonyBagnall

Enhancements

• [ENH] Shapelet transform changes (#857) @TonyBagnall

• [ENH] Speed up regression pipeline example (#860) @TonyBagnall

• [ENH] change collection conversions from convert_to to convert_collection (#838) @TonyBagnall

• [ENH] DL Refactor (#937) @TonyBagnall

Segmentation

Other

• [ENH] refactor annotation into segmentation and anomaly detection (#782) @TonyBagnall

Similarity Search

Enhancements

• [ENH] New similarity search module (#724) @TonyBagnall

Transformations

Bug Fixes

• [BUG] shapelet transform square root (#915) @TonyBagnall

Documentation

• [ENH] Introduce the “Tabular” output type tag and convert remaining Transformers to CollectionTransformers (#825) @TonyBagnall

Enhancements

• [ENH] BaseCollectionTransformer inherits from BaseCollectionEstimator and BaseTransformer (#709) @TonyBagnall

• [ENH] Shapelet transform changes (#857) @TonyBagnall

• [ENH] Support unequal length for rdst (#891) @baraline

• [ENH] remove nested_univ from Tabularizer (#897) @TonyBagnall

• [ENH] Add algorithm type tag to convolutional transformers (#930) @TonyBagnall

• [ENH] Add algorithm type tag to dictionary transformers (#931) @TonyBagnall

• [ENH] Add algorithm type tag to shapelet transformers (#932) @TonyBagnall

• [ENH] remove nested_univ from TSFresh test (#945) @TonyBagnall

• [ENH] Remove the TimeBinner transformer (#933) @TonyBagnall

• [ENH] Refactor polymorphic transformers out of series directory into root (#959) @TonyBagnall

Refactored

• [ENH] Refactor interval based transformers into their own subpackage (#916) @TonyBagnall

• [ENH] Refactor feature based transformers into their own subpackage (#929) @TonyBagnall

Other

• [ENH] Make MiniRocketMultivariateVariable conform to API (#839) @TonyBagnall

• [ENH] Adapt ShapeletTransform to work with unequal length (#820) @TonyBagnall

Other

Bug Fixes

• [BUG] Improve sweep.yaml comment trigger (#843) @wwzeng1

• [BUG] Update sweep.yaml spacing (#901) @wwzeng1

Documentation

• [DOC] network docstrings (#799) @TonyBagnall

• [DOC] Add a notebook to describe inheritance structure of base classes (#789) @TonyBagnall

• [ENH] scitype variable name and doc string purge (#798) @TonyBagnall

• [DOC] Update documentation files to allow easier local builds and raise sphinx bound (#796) @MatthewMiddlehurst

• [MNT] Proposed aeon deprecation policy (#723) @TonyBagnall

• [GOV] Code of Conduct revision (#806) @MatthewMiddlehurst

• [DOC] Update documentation to add workgroups (#842) @MatthewMiddlehurst

• [DOC] Updates remaining first level website pages (#847) @MatthewMiddlehurst

• [Sweep Rules] Remove commented code and improve instructive comments (#850) @sweep-ai[bot]

• [DOC] change n_dimensions to n_channels in all files (#918) @xiaopu222

• [DOC] Add warning to installation guide for some shells (#963) @MatthewMiddlehurst

• [DOC] Remove “Time series” from API titles (#980) @TonyBagnall

Enhancements

• [ENH] Fix pre-commit fails (#851) @TonyBagnall

• [ENH] Refactor X_inner_mtype (#859) @TonyBagnall

• [ENH] Set Draft to true in sweep.yaml (#900) @wwzeng1

• [ENH] Remove upper bounds for tslearn and pmdarima (#939) @MatthewMiddlehurst

• [ENH] Update numpy version (#935) @MatthewMiddlehurst

• [ENH] Update scikit-learn version (#940) @MatthewMiddlehurst

• [ENH] Update tensorflow (#912) @MatthewMiddlehurst

• [ENH] Refactor numpyflat to be numpy2D (#941) @TonyBagnall

• [ENH] Remove the alignment ADT from datatypes (#979) @TonyBagnall

Maintenance

• [MNT] Consolidate config files and add new pre-commit functionality (#340) @MatthewMiddlehurst

• [MNT] Comment out mypy pre-commit hook and run pre-commit on all files. Remove individual copyright notices (#801) @MatthewMiddlehurst

• [MNT] Update sweep.yaml (#821) @kevinlu1248

• [MNT] Add PR testing to test-nosoftdeps-full (#823) @MatthewMiddlehurst

• [MNT] Update Pull Request template (#829) @MatthewMiddlehurst

• [MNT] Add new sweep rules and make current rules more specific (#828) @MatthewMiddlehurst

• [MNT] Add auto label and welcoming comment workflow for PRs (#832) @MatthewMiddlehurst

• [MNT] Add new periodic tests and tidy workflows (#830) @MatthewMiddlehurst

• [MNT] Add workflow to update pre-commit hooks and make a PR (#831) @MatthewMiddlehurst

• [MNT] Improve coverage for utils/validation (#793) @TonyBagnall

• [MNT] Stop pre-commit bot on draft PR (#854) @MatthewMiddlehurst

• [MNT] Always create a pull request on pre-commit update even if full run fails (#863) @MatthewMiddlehurst

• [MNT] Allow manual run of certain workflows (#865) @MatthewMiddlehurst

• [MNT] Remove sweep rules and sandbox options (#874) @MatthewMiddlehurst

• [MNT] Always generate token for pre-commit version update workflow (#876) @MatthewMiddlehurst

• [MNT] Give more time for codecov workflow tests (#877) @MatthewMiddlehurst

• [MNT] Try to fix pre-commit auto update (#881) @MatthewMiddlehurst

• [MNT] Use app token for other GitHub workflows (#886) @MatthewMiddlehurst

• [MNT] Retry aeon install on failure for GitHub workflows (#885) @MatthewMiddlehurst

• [MNT] Setup file for Dependabot (automatic dependency updates) (#892) @MatthewMiddlehurst

• [MNT] Dependabot validation workflow (#903) @MatthewMiddlehurst

Refactored

• [MNT] Proposed aeon deprecation policy (#723) @TonyBagnall

Other

• [ENH] Bug and coverage for BaseCollectionTransformer (#808) @TonyBagnall

• [ENH] Refactor tags to remove scitype (#800) @TonyBagnall

• Improve docstrings formatting in deep learning classifiers (#814) @sweep-ai[bot]

• [ENH] Refactor scitype:y to y_input_type and y_inner_mtype to y_inner_type (#810) @TonyBagnall

• [GOV] New aeon governance document. (#195) @MatthewMiddlehurst

• Update sweep.yaml to disable rules and fix sandbox (#878) @kevinlu1248

Contributors

The following have contributed to this release through a collective 100 GitHub Pull Requests:

@akshatvishu, @baraline, @chrisholder, @dguijo, @hadifawaz1999, @kevinlu1248, @MatthewMiddlehurst, @PaulRabich, @TonyBagnall, @wwzeng1, @xiaopu222, @zy18811